17 A Framework for Dynamic Simulation of Interconnected …
623
Fig. 23 Temporal progression of the SMD and power input for a solution of the genetic algorithm
after 30 (top) and 65 (bottom) iterations
the dotted black line shows the target value of the SMD. In this case, the number of
parent and child solutions is 50 and the starting value of σ is set to 0.4. The optimizer
may vary the power input at 4 different time steps between 50 and 500 kW, which
leads to a total of 8 independent variables. The power input may be changed in a
time interval between 30 and 70 s.
In Fig. 23, it can be seen that the stepwise change of the input mass flow leads to a
distortion in the SMD of the outlet. To mitigate the impact of the change in the feed,
the power input of the crusher should be decreased in advance and after the change
in the feed the power input should be gradually increased until the target value of the
SMD is reached.
10 Approaches to Development of Data-Driven Flowsheet
Models
The abundance of data and the increasing computational capacities have led to a
wide use of data-driven models, like artificial neural networks (NNs), in different
fields of both science and industry. The main advantage of NNs is the efficient
623
Fig. 23 Temporal progression of the SMD and power input for a solution of the genetic algorithm
after 30 (top) and 65 (bottom) iterations
the dotted black line shows the target value of the SMD. In this case, the number of
parent and child solutions is 50 and the starting value of σ is set to 0.4. The optimizer
may vary the power input at 4 different time steps between 50 and 500 kW, which
leads to a total of 8 independent variables. The power input may be changed in a
time interval between 30 and 70 s.
In Fig. 23, it can be seen that the stepwise change of the input mass flow leads to a
distortion in the SMD of the outlet. To mitigate the impact of the change in the feed,
the power input of the crusher should be decreased in advance and after the change
in the feed the power input should be gradually increased until the target value of the
SMD is reached.
10 Approaches to Development of Data-Driven Flowsheet
Models
The abundance of data and the increasing computational capacities have led to a
wide use of data-driven models, like artificial neural networks (NNs), in different
fields of both science and industry. The main advantage of NNs is the efficient